CS598IQVIAClaims

This project uses DNN, RNN, and Auto-encoder models, and its variants to predict future medical costs for patients in the IQVIA database.

All the data generating code, and models are in the src directory.

Good starting points to look at the code are at the following links. Further text and documentation are provided in the ipynb notebooks.

Exploratory Data Analysos

https://github.com/balaksuiuc/CS598IQVIAClaims/blob/main/src/exploratory_data_analysis.py

Main DNN code

Main RNN+embedding code

https://github.com/balaksuiuc/CS598IQVIAClaims/blob/main/src/cost_prediction_RNN_embedding.ipynb

Autoencoding Experiments

Using Relu: https://htmlpreview.github.io/?https://github.com/balaksuiuc/CS598IQVIAClaims/blob/main/src/autoencoder_relu.html

Using tanh: https://htmlpreview.github.io/?https://github.com/balaksuiuc/CS598IQVIAClaims/blob/main/src/autoencoder_tanh.html

Linking to google drive

When running the codes, you will be asked to copy-paste the authorization for the raw/processed files in google drive.